Changing the Tone Mapping Method?

[rmi_]

Hi all,

I'm running Jellyfin 10.10.7 on UNRAID 7.1.0 (Linux Kernel 6.12.24), using the official docker container (from https://hub.docker.com/r/jellyfin/jellyfin). I'm using an AMD Ryzen 5 8500G with its integrated AMD Radeon 740M. I'm passing /dev/dri as a device to the container and Hardware De-/Encoding works fine (using VAAPI) until I enable Tone Mapping. With Tone Mapping enabled, Jellyfin doesn't play any HDR-content and immediately throws the error popup "Playback Error: Playback failed due to a fatal player error.". I do not enable VPP Tone mapping.

Log from Jellyfin: https://pastebin.com/Juz5v7Uv
Log from ffmpeg: https://pastebin.com/66AahkTJ

I guess it's trying to invoke OpenCL, but the docker container does not have ROCm. According to the Jellyfin Docs, there are two Tone-mapping Methods. How do I switch to Vulkan? I can't find any setting or documentation. Or is there a better way to do this?

Thanks in advance,
rmi_

[gnattu]

For AMD on Linux, we now only use the vulkan pipeline so the problem may be that in your container, there is no vulkan runtime at all. Can you share the ffmpeg log? for ffmpeg related issues the ffmpeg log would print the exact issue on what's wrong and you can find it in your server dashboard.

[rmi_]

Thanks for your answer. 

The ffmpeg-log is in the original post, or do you need something else (I might have gotten the wrong one)? It says "[AVHWDeviceContext @ 0x55fca80a0480] Failed to get number of OpenCL platforms: -1001", that's why I was guessing that it's OpelCL related.

[gnattu]

Sorry to miss that.

The server will log enviroment info during startup and you will see lines like the following:

VAAPI device {RenderNodePath} is AMD GPU, VAAPI device {RenderNodePath} supports Vulkan DRM modifier and VAAPI device {RenderNodePath} supports Vulkan DRM interop. What are the values of those lines? The server will only fallback to opencl when some of the vulkan extension is unavailable, or it failed to detect an AMD GPU is in place at all.

[rmi_]

Is this the part you're looking for?

Code:

[2025-05-06 07:35:34.282 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: Found ffmpeg version "7.0.2"
[2025-05-06 07:35:34.313 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: Available "decoders": ["libdav1d", "av1", "av1_cuvid", "av1_qsv", "h264", "h264_qsv", "h264_cuvid", "hevc", "hevc_qsv", "hevc_cuvid", "mpeg2video", "mpeg2_qsv", "mpeg2_cuvid", "mpeg4", "mpeg4_cuvid", "msmpeg4", "vc1_qsv", "vc1_cuvid", "vp8", "libvpx", "vp8_cuvid", "vp8_qsv", "vp9", "libvpx-vp9", "vp9_cuvid", "vp9_qsv", "aac", "ac3", "ac4", "dca", "flac", "mp3", "truehd"]
[2025-05-06 07:35:34.332 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: Available "encoders": ["libsvtav1", "av1_nvenc", "av1_qsv", "av1_amf", "av1_vaapi", "libx264", "h264_amf", "h264_nvenc", "h264_qsv", "h264_v4l2m2m", "h264_vaapi", "libx265", "hevc_amf", "hevc_nvenc", "hevc_qsv", "hevc_vaapi", "mjpeg_qsv", "mjpeg_vaapi", "aac", "libfdk_aac", "ac3", "alac", "dca", "flac", "libmp3lame", "libopus", "truehd", "libvorbis", "srt"]
[2025-05-06 07:35:34.349 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: Available filters: ["bwdif_cuda", "deinterlace_qsv", "deinterlace_vaapi", "flip_vulkan", "hwupload_cuda", "hwupload_vaapi", "libplacebo", "overlay_opencl", "overlay_qsv", "overlay_vaapi", "overlay_vulkan", "overlay_cuda", "procamp_vaapi", "scale_cuda", "scale_opencl", "scale_qsv", "scale_vaapi", "scale_vulkan", "tonemapx", "tonemap_cuda", "tonemap_opencl", "tonemap_vaapi", "transpose_cuda", "transpose_opencl", "transpose_vaapi", "transpose_vulkan", "vpp_qsv", "yadif_cuda", "zscale", "alphasrc"]
[2025-05-06 07:35:34.424 +02:00] [INF] [10] Emby.Server.Implementations.IO.LibraryMonitor: Watching directory "/data/movies"
[2025-05-06 07:35:34.457 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: Available hwaccel types: ["cuda", "vaapi", "qsv", "drm", "opencl", "vulkan"]
[2025-05-06 07:35:34.706 +02:00] [INF] [8] Emby.Server.Implementations.IO.LibraryMonitor: Watching directory "/data/tvshows"
[2025-05-06 07:35:36.780 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: VAAPI device "/dev/dri/renderD128" is AMD GPU
[2025-05-06 07:35:36.780 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: VAAPI device "/dev/dri/renderD128" supports Vulkan DRM modifier
[2025-05-06 07:35:36.780 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: VAAPI device "/dev/dri/renderD128" supports Vulkan DRM interop
[2025-05-06 07:35:36.780 +02:00] [INF] [1] MediaBrowser.MediaEncoding.Encoder.MediaEncoder: FFmpeg: "/usr/lib/jellyfin-ffmpeg/ffmpeg"
[2025-05-06 07:35:36.781 +02:00] [INF] [1] Emby.Server.Implementations.ApplicationHost: ServerId: "9ab007e51d524233a0902620e1903341"
[2025-05-06 07:35:36.781 +02:00] [INF] [1] Emby.Server.Implementations.ApplicationHost: Core startup complete

[gnattu]

Well the server is doing something I do not understand. From you ffmpeg log, the server failed detecting the vendor of the vappi driver which makes it initialize the generic path, but your server log indicates that the server detected an AMD GPU as vaapi vendor. Can you try just restart the container and see if it behaves normally?

[rmi_]

Thanks for your answers so far!

I usually shut down the whole system every night, and this has been happening for over a week. This error even survived the system upgrade from UNRAID 7.0.1 to 7.1.0, so it had lots of restarts 

I even tried the Jellyfin-Docker-Image from linuxserver.io instead of the official one. Same error.

Is there anything else I can do to debug this?